You are on page 1of 13

Environmental Management

DOI 10.1007/s00267-015-0562-5

Livestock Predation by Puma (Puma concolor) in the Highlands


of a Southeastern Brazilian Atlantic Forest
Francesca Belem Lopes Palmeira1,3 • Cristiano Trapé Trinca1,3 • Claudio Maluf Haddad2

Received: 4 February 2014 / Accepted: 8 June 2015


Ó Springer Science+Business Media New York 2015

Abstract We evaluated local opinion about reducing approximately 12 ± 5 animals, equivalent to 0.4 % of the
livestock losses to puma (Puma concolor) and the potential total livestock. Property elevation and distance from the
for conflict among livestock breeders inside a protected urban center were the main predictors of predation proba-
area in the highlands of a southeastern Brazilian Atlantic bility. This survey used a novel approach that has not been
forest. We also quantified the number and type of livestock addressed directly in other studies on livestock predation
losses, and determined if predation by puma was correlated and demonstrated that the high potential for conflict among
with property profile and landscape characteristics. We livestock breeders should be considered before imple-
conducted semistructured interviews with 42 livestock menting management actions.
breeders sampled in 36 rural properties. When asked how
to reduce predation, 33 % of livestock breeders refused to
answer, 26 % suggested improving livestock husbandry Keywords Attitudes  Conservation  Generalized linear
practices, 19 % stated that there was no appropriate action, model  Human–wildlife interaction  Potential for conflict
17 % favored removing the ‘‘problem’’ individual, and 5 % index
suggested killing the puma. Opinion on how to solve pre-
dation was independent of herd size and history of losses,
and was correlated with respondent age class. Older Introduction
respondents tended to suggest removing or killing pumas.
Attitudes toward predation represented high potential for The loss of livestock due to apex predators is a worldwide
conflict among livestock breeders who demonstrated high issue and involves both ecological and social factors of
discordance among responses. Horses were the most human–wildlife conflicts (Graham et al. 2005). Although
common prey (51 %), followed by cattle (28 %), sheep ‘‘human–wildlife conflict’’ is a widely accepted term to
(17 %), and goats (4 %); totaling 47 animals attacked describe negative interactions between wild animals and
between 2004 and 2007. Annual predation was people, perhaps a more appropriate term for the losses caused
by wildlife is human–wildlife relationship or human–wild-
life interaction. Indeed, aspects of human–wildlife conflict
& Francesca Belem Lopes Palmeira
yagouaroundi@yahoo.com.br; involve disagreements over how wildlife should be managed
http://www.jaguarjuruena.com (i.e., human–human conflict) (Peterson et al. 2010).
1
Big cats are responsible for attacks on livestock
Research Department, Reserva Brasil, Av Dr Silva Melo 520,
throughout their distribution, and the severity of conflict
Apto 606, Jd Taquaral, São Paulo, SP CEP 04675-010, Brazil
2
increases with felid body mass, which means higher levels
Department of Animal Science, ‘‘Luiz de Queiroz’’ College
of conflict for larger species such as caracals (Caracal
of Agriculture, University of Sao Paulo (ESALQ/USP), Av
Pádua Dias 11, Piracicaba, SP CEP 13418-900, Brazil caracal), cheetahs (Acinonyx jubatus), Eurasian lynx (Lynx
3 lynx), jaguars (Panthera onca), pumas (Puma concolor),
Present Address: Jaguar Juruena, R Joao Braz Alves, QD 04,
Lote 16, Japuranã, Nova Bandeirantes, MT CEP 78565-000, snow leopards (Uncia uncia), leopards (Panthera pardus),
Brazil lions (Panthera leo), and tigers (Panthera tigris) (Inskip

123
Environmental Management

and Zimmermann 2009). However, although there is communities and wildlife managers in the co-management
enormous potential for big cats to attack livestock, mor- of human–wildlife conflicts (Treves et al. 2006). Attitudes
tality by other causes (e.g., diseases, malnutrition, mal- toward predation may be affected by a wide range of
practice, and others) has a higher impact on livestock personal and sociocultural aspects (Marchini and Mac-
losses than predation (Oli et al. 1994; Mazzolli et al. 2002; donald 2012; Soto-Shoender and Main 2013) leading to
Palmeira et al. 2008). high disparity in opinions. Indeed, this conflicting opinion
Among the Brazilian biomes inhabited by jaguar and among stakeholders is a critical—yet hitherto largely
puma, the Atlantic Forest has suffered most from fragmen- unrecognized—first step to address for any successful
tation processes, with only 11–16 % of the original forested predator management and mitigation program.
area remaining (Ribeiro et al. 2009). This remaining area still Based on these observations, our goals were (1) to eval-
maintains high levels of biodiversity and endemism (Mit- uate the determinants of local opinion in relation to personal
termeier et al. 1999). Although 88 % of the original forest is profile, herd size, and history of livestock losses to puma and
completely depleted of four species of large mammals the potential for conflict among livestock breeders, (2) to
[jaguar, tapir (Tapirus terrestris), white-lipped peccary quantify the number and type of livestock losses to pumas,
(Tayassu pecari), and muriqui (Brachyteles arachnoides)], and (3) to determine if predation was correlated with prop-
55 % is considered unsuitable for them (Jorge et al. 2013). erty profiles and landscape characteristics in properties sur-
Even pumas, which appear to be more tolerant of forest loss rounding a protected area in the highlands of a southeastern
and fragmentation than jaguars (De Angelo et al. 2011), have Atlantic forest. The survey represents a novel approach that
been affected by loss of genetic variation (Miotto et al. 2011; has not been addressed directly in other studies on livestock
Castilho et al. 2012). There is evidence that hunting, and not predation. This is highly relevant as there is a current push to
lack of landscape connectivity, is a major threat for this analyze human–wildlife conflicts from a human dimensions
species (Castilho et al. 2011). Although illegal, the killing of and psychological perspective. Finding good predictors of
pumas as a result of livestock predation is carried out in the likelihood of livestock losses to pumas and attitudes
ranches across the biome (Guix 1997; Mazzolli et al. 2002; toward predation can help us develop preventive strategies to
Verdade and Campos 2004). Consequently, pumas exhibit mitigate the potential for conflict.
differences in activity patterns among areas with different
hunting pressures (Paviolo et al. 2009). When there is little
hunting and high tolerance by humans, puma populations Study Area
achieve high densities even in a modified landscape (Maz-
zolli 2010). Pumas consume a diet based on a broad range of This study was carried out surrounding the Campos do Jor-
prey and exhibit high behavioral plasticity in using anthro- dão State Park (CJSP), a protected area which is located
pogenic resources from the agricultural matrix (Magioli et al. between 22°370 –22°450 S and 45°240 –45°310 W, bordered by
2014), which should predispose the species to predating on São Paulo and Minas Gerais states, southeastern Brazil
livestock. (Fig. 1). It is part of the Serra da Mantiqueira Environmental
Public attitudes toward livestock losses due to big cats Protection Area (APA) composed of some parks and several
have become less tolerant, to the point where there are very private properties, totaling an area of 421,804.46 ha. The
few surviving populations outside of protected areas vegetation consists of high-altitude grasslands and Araucaria
worldwide (Jackson and Nowell 1996), thus increasing the forest (Araucaria angustifolia), which are typical ecosys-
conflict with wildlife conservation goals. To reduce hunt- tems of high and cold regions of the Atlantic forest biome.
ing as retaliation for livestock predation, several strategies CJSP covers an area of 8341 ha and the elevation ranges
have been discussed on a regional scale including predator from 1030 to 2007 m. The climate is subtropical (upland),
control (Linnell et al. 1997), improving livestock hus- with a dry winter and mild summer (cwb) as classified by
bandry practices (Ogada et al. 2003; Kissling et al. 2009), Köeppen. The average temperature varies from 27.2 °C in
paying financial compensation (Bulte and Rondeau 2005; the summer to -4.4 °C during the winter. Rainfall varies
Maclennan et al. 2009; Dickman et al. 2011), and devel- from 1500 to 2000 mm and occurs mostly during the spring
oping environmental education (Holmern et al. 2007). and summer (Seibert et al. 1975; Clauset and Soares 1999).
While there is a generally good understanding of the CJSP was one of the first regions to be settled in
dynamics of human–wildlife conflicts, this understanding southeastern Brazil. In 1975, the park’s management plan
can vary among sites depending on the local determinants. already reported changes to the vegetation resulting from
This could explain why there is no consistent or reliable anthropogenic activities in the region (e.g., selective log-
framework for managing predation (Graham et al. 2005). It ging, reforestation with exotic conifers, extensive pastures,
should be necessary to combine baseline research, partic- management of fields with fire and disorderly occupation;
ipatory planning, and monitoring to engage affected Seibert et al. 1975).

123
Environmental Management

Fig. 1 Location of respondents (n = 42) and reported predation on livestock in the Campos do Jordão State Park (CJSP) and Serra da
Mantiqueira Environmental Protection Area (APA), southeastern Brazil

Methods livestock breeders (from May to August 2007) in an


attempt to establish trust between the interviewer and the
Data Collection interviewee, and provide more consistent data collection
(Oli et al. 1994; Palmeira and Barrella 2007).
From July 2006 to August 2007, we undertook nine field We selected all breeders who reared livestock in the
expeditions, totaling 80 days of sampling effort. We con- park and surrounding properties, totaling 42 respondents
ducted interviews 10 months after first contact with sampled in 36 properties. We only interviewed one

123
Environmental Management

respondent per property (n = 35), with the exception of the dimensional contingency tables considering the absolute
park (n = 1) where there were three employees responsible frequencies (n) for each response (improving husbandry,
for livestock kept in the area. We also interviewed four removing pumas, killing pumas, no appropriate action, and
herders who had no property, but kept their livestock loose would not answer) in relation to the age class of the
in the park or on leased pastures of the surrounding prop- respondent, the herd size, and the history of losses
erties. We conducted one interview per interviewee, lasting (Table 2).
an average of 26 ± 13 min, with a minimum of 9 and To assess conflict among respondents’ opinions, we
maximum of 68 min. The interviews were held face to face adopted the potential for conflict index (PCI2), which is a
at the main house of each property and conducted in Por- descriptive statistical method to facilitate the understanding
tuguese, which is the native language of interviewees and of results in human dimensions research (Manfredo et al.
interviewers. 2003; Vaske et al. 2006, 2010). PCI2 ranges from 0
We divided the questionnaire into three sections: (1) (minimum potential for conflict) to 1 (maximum potential
profiles of properties and respondents; (2) livestock pre- for conflict). The maximum potential for conflict
dation; and (3) local opinion (Table 1). During sections (1) (PCI2 = 1) occurs when 50 % agree, 50 % disagree, and
and (2), we used structured questions with precise wording 0 % are neutral. In contrast, no potential for conflict
and fixed response categories. To avoid influencing the (PCI2 = 0) occurs when 100 % agree, disagree, or are
replies in section (3), we asked semistructured questions, neutral (Vaske et al. 2010). To calculate the PCI2, we
which enabled respondents to express their opinions with reorganized five dichotomous questions (yes, no, or nei-
spontaneous responses. To obtain the property profile, we ther) of the semistructured questionnaire (Table 2) into a
recorded each property’s purpose and size (ha) and used bipolar scale, with scale widths from -1 (minimum scale
GPS to extract the geographic location (X and Y) of the value) to 1 (maximum scale value) and the distance func-
sites where livestock were kept. To obtain the respondent tion D2, which includes a neutral value. Power function
profiles, we recorded each respondent’s age, the length of and number of repetitions for the simulation were accord-
residence in the region, place of birth, and their main ing to default values, 1 and 400, respectively. We com-
economic activity. To assess the damage caused by pre- puted the frequency (n) and scale mean for each variable
dation, we asked questions regarding the number and type among each group of respondents. Therefore, we compared
of livestock reared by respondents, husbandry practices, the statistical differences (d) among respondents with and
number and type of livestock attacked, and whether there without prior history of l losses, where d [ 1.96 is signif-
had been sightings of predated animals. If carcasses were icant at P \ 0.05 (Vaske et al. 2006). We displayed the
found, we asked for further details including the parts results as bubble graphs to illustrate dispersion, central
consumed, age and sex, date and locale, what happened tendency and form simultaneously, where large bubbles
after the attack, and other observations. To determine if the reflect more potential conflict and small indicate less
livestock death was due to attacks by pumas, respondents (Manfredo et al. 2003). To calculate the PCI2, we used the
usually examined the carcass or injured animal for internal standalone online version (PCI2sa) and plotted the bubble
and external injuries (bites and scratches), especially in the graph using Microsoft PowerPoint templates (http://war
skull, neck, and limbs. Pumas generally bite prey at the nercnr.colostate.edu/^jerryv/PCI2/index.htm).
throat, resulting in death by asphyxiation. Occasionally, To reduce potential bias due to differences in memory of
they can cover the carcass with leaves and other materials respondents, we only quantified the frequencies (n), per-
to hide it from other predators (Hoogesteijn 2005). How- centages (%), mean (x), and standard deviation (SD) of
ever, we were very confident in the respondents’ ability to livestock attacked more recently (2004–2007) as per sim-
distinguish attacks by puma, we discarded unreliable ilar surveys (Oli et al. 1994; Palmeira and Barrella 2007;
reports to eliminate any biases in our study. To evaluate Soto-Shoender and Giuliano 2011).
local opinion, we recorded respondents’ suggestions for To determine when livestock predation (response vari-
reducing losses and their feelings regarding the presence of able) occurred in relation to some possible explanatory
pumas in the area. variables, we used a logistic regression model [generalized
linear model (GLM)] for a binomial distribution (Soto-
Data Analysis Shoender and Giuliano 2011; Teichman et al. 2013). We
created a binary history of predation (1 = presence and
To evaluate whether opinion on how to solve predation was 0 = absence) for each respondent (n = 42) considering all
associated with factors such as respondent profile, herd occurrences (even before 2004). As explanatory variables
size, and history of losses, we used Pearson’s v2 analysis (Table 2), we used the property profile (area and herd sizes
(v2; P \ 0.05), which is especially suited for nominal data per respondent) and extracted landscape covariates such as
(Zar 2010). To calculate the test, we arranged two- elevation, tree cover, and Euclidian distance from the

123
Environmental Management

Table 1 Semistructured questionnaire used to interview livestock breeders about puma predation on their livestock

Property and respondent profile

1. Name of property owner:

2. Property size (ha):

3. Main house (UTM): X/Y Elevation (m): GPS ID:

4. Property purposes: ( ) subsistence ( ) holiday home ( ) tourism ( ) animal ( ) crop ( ) others?

5. Name of respondent:

6. Age:

7. Place of birth:

8. Time of residence at the property:

9. Main activity:

10. Position: ( ) owner ( ) employee ( ) others?

Livestock predation

11. Number and type of livestock reared:

( ) cattle ( ) horses ( ) goats ( ) sheep ( ) pigs ( ) poultry ( ) dogs ( ) others?

12. Husbandry practices:

( ) extensive ( ) nightly confinement ( ) seasonal confinement ( ) total confinement ( ) others?

13. Did already have livestock losses due to puma? ( ) Y ( ) N

14. Number and type of livestock attacked by puma:

Date Livestock Age Gender 1234* Parts consumed Kill site (X/Y) Elevation (m)

* Livestock conditions: (1) injured (2) killed/covered (3) killed/uncovered (4) others?

15. What did you do after the attack? ( ) changed husbandry practices ( ) nothing ( ) others?

16. Did you see the puma? ( ) Y/Date/Site? ( ) N

17. Do you know someone or someplace that had livestock losses duo to puma? ( ) Y/Who/Where? ( )

Perception

18. What kind of big cats exist on the property? ( ) puma ( ) jaguar ( ) others?

19. Do you know the puma? (show photo) ( )Y( )N

20. What kind of ecological role of puma or for what puma exist? ( ) preserve/control ( ) prettify ( ) no

role ( ) don’t know ( ) others?

21. What do you think of puma close to where you live? ( ) like ( ) dislike ( ) indifferent ( ) don’t

know ( ) others?

123
Environmental Management

Table 1 continued

Perception

22. Where the most suitable place for puma to live? ( ) CJSP ( ) forest ( ) zoo ( ) don’t know ( )

others?

23. Puma poses a threat to livestock? ( ) Y/Why? ( ) N

24. Puma poses a threat to humans? ( ) Y/Why? ( ) N

25. Could you suggest how to solve livestock losses by puma predation? ( ) improve husbandry practices

( ) financial compensation ( ) remove the problem puma ( ) kill the puma ( ) don’t know ( ) others?

26. Who should bear responsibility for the damage caused by puma predation ? ( ) government ( )

livestock owner ( ) don’t know ( ) others?

27. Do you think that is necessary to control livestock predation? ( ) Y ( ) N

28. Would you change husbandry practices to reduce losses due to puma? ( ) Y ( ) N

29. Would you accept suggestions to prevent losses? ( ) Y ( ) N

30. Would like to receive the results and/or other materials from this survey? ( ) Y/Address: ( ) N

largest urban center in the study area (Municipality of with elevation. Variable elevation was selected over per-
Campos do Jordão) using ArcGis 9.3 (ESRI 2008). To cent tree cover because there is a considerable number of
avoid colinearity between the explanatory variables, we properties located in the high-altitude grasslands where the
calculated a Pearson’s correlation coefficient (r; P \ 0.05) percent of tree cover is lower or absent and livestock
(Zar 2010) before adding them to the model. After this predation occurs as well. Then, the explanatory variables
step, we excluded the percent of tree cover from the models were standardized to a mean of zero and unit variance. We
because it was associated positively (r = 0.47; P \ 0.01) tested the overall model fit regarding a subset of regression

Table 2 Variables selected for


Variable description Range Value Analyses
the analyses
Tree cover (%) 27–75 Continuous GLM
Elevation (m) 1012–1985 Continuous GLM
Distance from the urban center (m) 6771–29,850 Continuous GLM
Area size (ha) 1–8341 Continuous GLM
Herd size 1–960 Continuous GLM
Herd size \20 Categorical v2
21–60
[61
Age class 18–35 Categorical v2
36–53
[54
History of livestock attacks Yes or no Categorical v2
Puma poses a threat to livestock Agree, disagree or neutral Scale (-1 to 1) PCI2
Puma poses a threat to humans Agree, disagree or neutral Scale (-1 to 1) PCI2
It is necessary to control depredation Agree, disagree or neutral Scale (-1 to 1) PCI2
Would change husbandry practices Agree, disagree or neutral Scale (-1 to 1) PCI2
Would accept suggestions to prevent losses Agree, disagree or neutral Scale (-1 to 1) PCI2
GLM generalized linear models, v2 Chi square, PCI2 potential for conflict index

123
Environmental Management

parameters using the maximum likelihood estimate and Local Opinion


Akaike’s information criterion (AIC). We developed a
priori a set of candidate models including a global model When respondents were asked what kind of big cats exist
with all parameters (K = 5) and with the link function on the property, the puma was the species most frequently
‘logit.’ We used the ‘‘corrected’’ AIC (AICc) a second- cited (74 %) followed by the ocelot (Leopardus pardalis;
order AIC for small samples, the top delta AICc (DAICc 21 %), the jaguar (12 %), and other small felids (Leopar-
proximity to zero) and greater AICc weights (w) to select dus spp.; 5 %). When asked about the ecological role of
the best-fit model (Burnham and Anderson 2002). We used puma, 55 % would not answer, 19 % said preserving the
the inverse link function ‘plogis’ to interpret the parameters ecosystem or controlling native prey, 19 % said attacking
and residuals deviance (variance and mean). We also livestock or causing damage, and 7 % said no role. In
examined overdispersion (/) from the global model where relation to what they thought of puma close to where they
c-hat [1 (dispersion parameter) indicates more variance live, 38 % dislike or feel fear, 38 % are indifferent, 19 %
than would be expected. Values much higher than 1 ([4) like, and 5 % would not answer. When we asked where
probably indicate a lack-of-fit (Burnham and Anderson would be the most suitable place for puma to live, 62 %
2002; Mazerolle 2013). When variances are greater than 1, said the forest or the jungle, 14 % would not answer, 12 %
the Quasi-likelihood (QAICc) models should adjust the said the CJSP, 7 % said no place, and 5 % said the zoo.
expected variance to account for / (Bolker 2008). To test When we asked who should bear responsibility for the
the GLM, Pearson’s correlation, and v2 analyses, we used damage caused by predation (i.e., pay for the damages),
the package ‘stats’ in the R environment version 3.0.2 33 % thought the government, 31 % said the livestock
(2013-09-25; R Core Team 2013). Also, to rank the models owner, 24 % said no one, and 12 % would not answer.
based on AIC, we used the package ‘AICcmodavg’ Thirty-three percent of the respondents would not
(Mazerolle 2013) and plotted the best-fit model in the answer the question about how to solve livestock losses,
‘graphics’ (R Core Team 2013). 26 % cited that improving livestock husbandry practices
would be the best solution, 19 % said no appropriate
action, 17 % said removing the ‘‘problem’’ individual from
the area, and 5 % said killing pumas. Note that these last
Results three suggestions were cited after 24 % of respondents
stated ‘‘cannot kill.’’ Specific husbandry practices were
Respondent Profile mentioned by respondents, for example, monitoring the
border, prevention, protection, fences, and confines, using
The 36 visited properties covered a total area of collar bells around the necks of domestic animals, intro-
17,380.40 ha. Sixty-four percent of them ranged in size ducing horned bulls into the herds, and maintaining wild
from 1 to 84 ha and 30.5 % from 144 to 780 ha. Only populations of preferred prey species that pumas like to eat.
5.5 % of properties covered large areas, in particular, the The opinion on how to solve the predation problem was
CJSP totaling 8341 ha and a sheep ranch totaling approx- independent of herd size (v2 = 2.53, df = 4, P = 0.63)
imately 4800 ha. Property purposes were subsistence and history of losses (v2 = 1.65, df = 2, P = 0.43), and
agriculture (30.5 %), followed by holiday homes (14 %), correlated with respondent age class (v2 = 13.23, df = 4,
holiday homes with crop or animal production (14 %), P = 0.01). Respondents up to 35 years tended to suggest
tourism with animal production (14 %), animal production improving livestock husbandry practices, whereas respon-
(11 %), crop production (5.5 %), and other activities (the dents older than 54 years tended to suggest removing the
CJSP, the Brazilian army, a monastery, and a real estate ‘‘problem’’ individual from the area or killing pumas. Most
development; 11 %). of the intermediate age class between 36 and 53 years
The 42 respondents were 93 % male and 7 % female. refused to answer questions on how to solve the predation.
The mean age was 45 ± 15 years, with a minimum of 18 The PCI2 for attitudes of the respondents demonstrated
and maximum of 72 years. The average time of residence high discordance among dichotomous responses (Fig. 2).
on the current property was 31 ± 21 years, with a mini- The questions ‘‘do pumas pose a threat to livestock?’’, ‘‘do
mum of one and maximum of 65 years. About 90 % of the pumas pose a threat to humans?’’ and ‘‘would you change
respondents were born in municipalities inserted in the husbandry practices?’’ represented more potential conflict
APA. The main activity of the respondents was classified due to the extreme values of favorable (*50 %) and
into general service (homemaker and housekeeper; 43 %), unfavorable (*50 %) responses. To the questions ‘‘is it
animal husbandry (17 %), technical service (14 %), necessary to control predation?’’ and ‘‘would you accept
administrative service (9.5 %), smallholder (7 %), and suggestions to prevent losses?’’, the responses were more
others (retired, military, monk, and public servant; 9.5 %). strongly in agreement (*71 %) than disagreement and

123
Environmental Management

Fig. 2 Potential for conflict


index (PCI2) for attitudes
among livestock breeders
toward livestock predation by
pumas. PCI2 ranges from 0 (no
conflict) to 1 (maximum
conflict) and is represented by
bubble size. Small bubbles
indicate less potential conflict
among respondents and large
reflect more conflict

reflected less potential conflict. There was no difference The reported losses were separated into three groups: the
among respondents with and without history of losses presence of dead livestock (60 %), disappearance of ani-
because the differences in observed PCI2 values were not mals (36 %), and injured animals (4 %). According to
statistically significant (d \ 1.96; P [ 0.05). reports, the carcasses bore marks typical of puma attacks,
such as scratches on the neck and back. The parts most
Livestock Predation commonly eaten were head, neck, chest, ribs, groin, back,
and leg. It was also very common for the predator to drink
The most numerous livestock were sheep, cattle, poultry, the blood and separate the viscera to consume the liver and
and horses, followed by pigs and goats (Table 3). The heart. Of the total livestock found dead, 17 were uncovered
distribution of the type and number of livestock by and 11 were covered by foliage. The owners only consid-
respondent was not uniform. Although many kept horses, ered disappearances as being predation events when a loss
poultry, and cattle, few respondents kept pigs, sheep, and occurred simultaneously in neighboring properties and the
goats. The average number of livestock per respondent was animal remains were visually examined.
61 ± 147, with a minimum of one and maximum of 960 Half of the respondents (50 %) reported that they had
animals. Extensive farming was used most frequently, lost some type of livestock due to pumas. Considering only
followed by semiextensive (e.g., nightly confinement), and the most recent events (2004–2007), less than half (38 %)
intensive (total confinement). When necessary, 26 % of suffered losses (Fig. 1). The average livestock loss reported
respondents used more than one type of husbandry in per respondent was 1 ± 2 (n = 42), while the average only
accordance with the type of livestock, time of year, for respondents who had losses (n = 16) was 3 ± 3, with a
financial need, and other factors. minimum of 1 and maximum of 11 animals. Following the
respondents who have livestock losses, nine owners did
nothing, four changed their herd husbandry practices, and
Table 3 Number of losses attributed to pumas from 2004 to 2007
and herd size estimated in 2007 three buried the carcasses of the animals killed.
Predation on horses was most commonly reported
Livestock 2004 2005 2006 2007 Total Herd size (2007)
(51 %), followed by cattle (28 %), sheep (17 %), and goats
Horses 5 4 10 5 24 279 (4 %), totaling 47 animals attacked (Table 3). Mean annual
Cattle 1 1 3 8 13 695 predation was approximately 12 ± 5 animals attacked,
Sheep 1 2 3 2 8 860 equivalent to 0.4 % of the total livestock at the properties.
Goats 1 0 1 0 2 4 Of the animals attacked, 57 % were males, 30 % females,
Pigs 0 0 0 0 0 36 and the gender of 13 % could not be determined, often
Poultry 0 0 0 0 0 689 because they were newborn animals without sexual iden-
Total 8 7 17 15 47 2563 tification. Regarding age group, only 32 % animals had
their age estimated, eight of which were up to 1 year, four

123
Environmental Management

Table 4 Generalized linear


Candidate Models K AICc DAICc w
models (GLM) for determining
the probability of livestock Landscape (elevation ? distance from the urban center) 3 56.51 0.00 0.44
predation on the site covariates
Landscape (elevation ? distance from the urban center) ? herd size 4 58.14 1.63 0.19
Landscape (elevation ? distance from the urban center) ? area size 4 58.56 2.05 0.16
Constant model (*parameters) 1 60.32 3.81 0.06
Global model (all parameters) 5 60.41 3.89 0.06
Property profile (area ? herd sizes) ? elevation 4 61.14 4.63 0.04
Property profile (area ? herd sizes) 3 62.10 5.59 0.03
Property profile (area ? herd sizes) ? distance from the urban center 4 62.90 6.39 0.02
K number of parameters, AICc ‘‘corrected’’ Akaike’s information criterion for small samples, DAICc delta
AICc differences, w Akaike weights

elevations was demonstrated using the best-fit model


(Fig. 3).

Discussion

In our interviews, there was a high non-response rate,


especially for the question on how to solve livestock losses.
As demonstrated in the results, many respondents stated
‘‘cannot kill’’ before answering the question, suggesting
that they were fearful of being reported to the authorities if
they suggested that problem pumas should be killed, even
after we promised anonymity and confidentiality on the
questionnaire application. The principal implication of the
non-responses was the lack of consensus among livestock
breeders and an increase in the potential for conflict. With
Fig. 3 Probability of livestock predation in relation to elevation and the exception of respondent age, the opinions on how to
distance from the urban center of the best-fit generalized linear model solve predation by pumas were independent of the
(GLM)
respondent profile, the characteristics of the property, or
the occurrence of losses. In general, opinions regarding
2 years, and three over 10 years old. For almost half of the livestock predation by pumas and jaguars do not seem to be
losses (45 %), the exact time when the attacks occurred associated with the respondent profile, herd size, number of
could not be identified. The remainder occurred during the losses (Palmeira and Barrella 2007), or size of the property
rainy season between spring and summer (September– (Conforti and Azevedo 2003; Zimmermann et al. 2005).
January; 34 %) and the dry season that precedes the winter Occasionally, the opinion is associated with the number of
(April, May, and June; 21 %). livestock held by respondents (Conforti and Azevedo
Based on the top DAICc (proximity to zero) and w, the 2003), the history of attacks (Soto-Shoender and Main
supported model indicated that predation was expected to 2013), and the age of respondents (Zimmermann et al.
be higher in areas far from the urban center and at high 2005).
elevations (Table 4). There was a moderate / in the global The majority of attitudes toward predation suggested
model because the variance was slightly greater than 1 (c- high potential for conflict among the respondents in the
hat = 1.31), and for this reason, we opted for AICc rather survey. The questions ‘‘do pumas pose a threat to livestock
than QAICc. The selected model had K = 3, or humans?’’ and ‘‘would you change husbandry prac-
DAICc = 0.00, w = 0.44. The model also had good pre- tices?’’ represented higher potential for conflict because of
dictive power (mean P = 0.46, range 0.16–0.84), and the lack of consensus among livestock breeders. Puma
explained 49.88 % of the deviance. Area and herd sizes attacks on humans are rare in Brazil, and despite there
were weak predictors of the probability of livestock pre- being no record of fatal attacks in the study area, half of
dation in the models. The probability of predation occur- livestock breeders reported that the species poses a threat to
ring in areas far from the urban center and at high humans as well as to their livestock. In contrast, livestock

123
Environmental Management

breeders tended to be in agreement with the questions ‘‘is it associated with seasonal birth of livestock (Palmeira et al.
necessary to control predation?’’, representing a moderate 2008) or the availability of alternative prey at different
conflict level, and ‘‘would you accept suggestions to pre- times of the year (Polisar et al. 2003; Patterson et al. 2004;
vent losses?’’ being the lowest conflict level. There was a Kolowski and Holekamp 2006).
discord between high consensus to accept suggestions to Although extensive management of livestock was the
prevent losses and little consensus to change husbandry most frequently used management method, respondents
practices. The results of the potential for conflict should be employed more than one type of husbandry simultaneously
used to guide implementation of mitigation measures and depending on need. Synchronized husbandry seems to be a
demonstrate a critical need to obtain a consensus of opinion very common practice among livestock breeders (Palmeira
before implementing management actions. and Barrella 2007), which limits our ability to investigate
Predation was low and represented only 0.4 % of the further the relationship between the livestock husbandry
livestock held in the study area. Other surveys have also practices adopted and the number of losses. Nevertheless,
shown that the number of losses attributable to predator some studies have suggested that predation is strongly
attacks is relatively low and generally does not exceed 3 % associated with husbandry practices adopted by farmers,
of the total stock (Graham et al. 2005). We believe that and that changes could reduce some of the attacks (Polisar
although pumas might not cause all livestock disappear- et al. 2003; Ogada et al. 2003; Wilson et al. 2006; Li et al.
ances, these reported disappearances play an important role 2013). A simulation using three different management
in human–wildlife interactions as respondents consider scenarios concluded that adopting a rotation of seasonal
predators responsible. Commonly, livestock disappear- breeding could reduce the exposure of animals to pumas
ances with no carcasses being found have also represented (Kissling et al. 2009). In other studies, the losses were not
a large proportion of losses reported by herders and tend to associated with the type of husbandry adopted (Graham
be blamed on predators (Palmeira and Barrella 2007; et al. 2005), the size or structure of the pens, or the number
Rosas-Rosas et al. 2008). of guard dogs or herders (Hemson et al. 2009).
Horses were the most frequently attacked livestock, Elevation and distance from the urban center were
and although well distributed spatially, they did not rep- important predictors of the supported model. Areas far
resent the most abundant livestock in the study area, from the urban center and at high elevations are more
which indicates that predation was selective and influ- likely to suffer losses to pumas because of the habitat
enced by the preference of the predator. This fact has suitability for the species. Part of this population is found
been well reported among wild carnivores that attack only in original habitat remnants above 800 m (Mazzolli
certain types of livestock more than expected (Namgail 1993). In British Columbia, cougar–human conflicts were
et al. 2007; Palmeira and Barrella 2007; Azevedo 2008; more likely to occur close to roads and at intermediate
Zarco-González et al. 2012). Pumas attacking horses are elevations (Teichman et al. 2013). In Mexico, predation
seldom reported in the literature; they usually attack foals was correlated with vegetation type and distances to per-
and young horses but rarely adult horses (Graipel et al. manent water and roads (Rosas-Rosas et al. 2010). In
2004; Shaw et al. 2007; Gray et al. 2008). When large Guatemala, forest cover, distance to forest, bodies of water
prey like buffalo (Azevedo et al. 2010) or cattle are and human settlements were also landscape predictors of
attacked, these are generally young or newborn individ- the probability of livestock predation (Soto-Shoender and
uals (Polisar et al. 2003; Palmeira et al. 2008). This trend Giuliano 2011).
was also found in our study as the horses and cattle that Despite the low rates of predation, some preventive
were attacked were mostly juveniles. Permanent confine- strategies must be addressed to reduce livestock losses.
ment of juveniles and nightly confinement of adults Managing problem animals such as killing or re-locating
should be implemented as alternative husbandry practices offending pumas might not be possible because the
to mitigate this predation. Atlantic forest is an endangered biome where there is
The livestock losses with a confirmed date occurred evidence of a genetic bottleneck for pumas (Miotto et al.
during two distinct periods: in the rainy season between 2011; Castilho et al. 2012). Even if the problem animal is
spring and summer and in the dry season that precedes detected, the efforts to capture are enormous and translo-
winter. The occurrence of livestock predation by carnivores cation appears to have a low success rate (Linnell et al.
in specific periods seems to be very common and has been 1997). In the same way, removing problem individuals will
observed both in winter (Palmeira and Barrella 2007) and not work in multi-use landscapes where livestock are dis-
in summer, and during the wet season or dry season, persed throughout the same areas in which carnivores are
depending on the region (Patterson et al. 2004; Kolowski to be conserved (Linnell et al. 1999). Also, financial
and Holekamp 2006; Palmeira et al. 2008; Li et al. 2013). compensation for those suffering losses is not a solution to
These peaks of predation in certain seasons may be the problem of conserving large carnivores (Dickman et al.

123
Environmental Management

2011) and can in fact decrease the efforts of livestock Acknowledgments Thanks to the Horto Florestal (Campos do
breeders to prevent losses and exacerbate conflicts with Jordão State Park), especially Marco Antonio P. Marcondes and
Waldir Joel de Andrade, and the Instituto Florestal (Process SMA N
wildlife (Bulte and Rondeau 2005). Therefore, more 42.769/2006) for institutional support and permission to conduct this
financial incentives are necessary for husbandry practices survey. Thanks to all the livestock breeders who kindly agreed to
that reduce probability of attacks, such as nightly con- participate in the interviews, especially Seu Jorge, Pedro Henrique de
finement for horses, sheep and goats, and keeping calves Oliveira, Fabinho, and Nelson M. Marcelino for their field assistance
in data collection. We thank the Environmental/Forest Police and Fire
away from the sites inhabited by big cats. Financial Department at Campos do Jordão for information provided, as well as
incentives could also be applied for intensive husbandry Pedro Paulo Filho, Arthur Bicudo, Antonio Scafuto Neto, Orlando
practices that improve livestock production as well as Paggiaro, Mr. Roberto (Romalar), Fotótica Shimazu, Fernando, Tião
finding sustainable, wildlife friendly and compatible and Cali for donation of photos and other materials. We thank our
parents, Ana Maria T. Trinca, Walter Trinca, and Benvinda B. Lopes
alternatives to supplementing their income. for the infrastructure provided and complete financial support of
Developing environmental education and offering fieldwork activities. Thanks to Ronaldo G. Morato for a reference
technical support for husbandry practices should be further provided and Renata A. Miotto for helping us to find the target
developed, because most respondents reported that they journal. We are very grateful for helpful comments and suggestions
from reviewers, Heather Aslin and José R. Soto-Shoender, and the
would be able to accept suggestions to prevent livestock editors-in-chief.
losses, which represents the lowest potential for conflict.
Actions of environmental education should be compre-
hensive and provide knowledge regarding a broad range of References
topics on biology, ecology, and conservation (e.g., eco-
logical role of predators and natural prey, their conserva- Azevedo FCC (2008) Food habits and livestock depredation of
tion significance, and intrinsic value). These actions could sympatric jaguars and pumas in the Iguaçú National Park area,
south Brazil. Biotropica 40:494–500
be performed by a trained team of professionals in social Azevedo FCC, Concone HVB, Pires-da-Silva A, Verdade LM (2010)
and biological sciences and directed to children, key-in- Puma (Puma concolor) predation on a water buffalo (Bubalus
formants, and information disseminators (e.g., teachers, bubalis). Mammalia 74:431–432
monitors and park guards, tourist guides, and volunteers). Bolker BM (2008) Ecological models and data in R. Princeton
University Press, New Jersey
The activities should occur in the field or classroom using Bulte EH, Rondeau D (2005) Why compensating wildlife damages
some concepts of the constructivism theory where humans may be bad for conservation. J Wildl Manag 69:14–19
can be active learners through their experiences. A specific Burnham KP, Anderson DR (2002) Model selection and multi-model
environmental education program must be directed to inference: a practical information-theoretic approach, 2nd edn.
Springer, New York
livestock breeders incorporating technical information on Castilho CS, Marins-Sá LG, Benedet RC, Freitas TO (2011)
alternative husbandry practices to improve animal pro- Landscape genetics of mountain lions (Puma concolor) in
duction and, consequently, reduce the risk of livestock southern Brazil. Mamm Biol 76:476–483
losses and the possible retaliatory hunting of predators. Castilho CS, Marins-Sá LG, Benedet RC, Freitas TR (2012) Genetic
structure and conservation of Mountain Lions in the south-
Also, it should be necessary to monitor losses to the Brazilian Atlantic rain forest. Genet Mol Biol 35:65–73
properties inserted in a landscape located in areas with high Clauset R, Soares D (1999) Paisagem paulista: áreas protegidas.
probability of predation, particularly those properties situ- Empresa das Artes, São Paulo
ated far from urban centers and at high elevations. Conforti VL, Azevedo FCC (2003) Local perceptions of jaguars
(Panthera onca) and pumas (Puma concolor) in the Iguaçú
This study’s recommendations were devised by con- National Park area, south Brazil. Biol Conserv 111:215–221
servation biologists in collaboration with an agronomist De Angelo C, Paviolo A, Di Bitetti M (2011) Differential impact of
specialized in animal production and should be directed to landscape transformation on pumas (Puma concolor) and jaguars
a broad range of stakeholders such as livestock breeders, (Panthera onca) in the Upper Paraná Atlantic forest. Divers
Distrib 17:422–436
wildlife conservationists, and political or decision-makers. Dickman AJ, Macdonald EA, Macdonald DW (2011) A review of
It is important to emphasize that our inference to a broader financial instruments to pay for predator conservation and encour-
and more general context and scope comes from a small age human–carnivore coexistence. PNAS 108:13937–13944
sample size, and the results and conclusions only apply to Environmental Systems Research Institute (ESRI) (2008) ArcGis
Desktop: Release 9.3.1. Environmental Systems Research Insti-
the study area. Because predictors of livestock predation tute, Redlands
should vary among sites, the choice of management actions Graham K, Beckerman AP, Thirgood S (2005) Human–predator–prey
would also depend on the region. This survey represented conflicts: ecological correlates, prey losses and patterns of
the first attempt in the study of livestock predation to management. Biol Conserv 122:159–171
Graipel ME, Ghizoni IV Jr, Mazzolli M (2004) Selvageria ou carência
evaluate potential for conflict among livestock breeders nutricional? Cienc Hoje 35:62–65
regarding predators and it is the main contribution and Gray M, Spencer J Jr, Thain D (2008) Live trapping and monitoring
novelty of this study. mountain lion movements with in a feral horse population in

123
Environmental Management

storey county, Nevada, 2005–2007. In: Timm RM, Madon MB conservation in a human-dominated landscape. Conserv Genet
(eds) Proceedings of the 23rd Vertebrate Pestide Conference, 12:1447–1455
University of California, Davis, pp 140–144 Mittermeier RA, Myers N, Gil PR, Mittermeier CG (1999) Hotspots:
Guix JC (1997) Cat communities in six areas of the state of São earth’s biologically richest and most endangered terrestrial
Paulo, southeastern Brazil, with observations on their feeding ecoregions. CEMAX and Conservation International (CI), Mex-
habitats. Grupo Estud Ecol Ser Doc 5:16–38 ico City
Hemson G, Maclennan S, Mills G, Johnson P, Macdonald D (2009) Namgail T, Fox JL, Bhatngar YV (2007) Carnivore-caused livestock
Community, lions, livestock and money: a spatial and social mortality in Trans-Himalaya. Environ Manag 39:490–496
analysis of attitudes to wildlife and the conservation value of Ogada MO, Woodroffe R, Oguge NO, Frank LG (2003) Limiting
tourism in a human–carnivore conflict in Botswana. Biol depredation by African carnivores: the role of livestock
Conserv 142:2718–2725 husbandry. Conserv Biol 17:1521–1530
Holmern T, Nyahongo J, Røskaft E (2007) Livestock loss caused by Oli MK, Taylor IR, Rodgers ME (1994) Snow leopard Panthera uncia
predators outside the Serengeti National Park, Tanzania. Biol predation of livestock: an assessment of local perceptions in the
Conserv 135:534–542 Annapurna Conservation Area, Nepal. Biol Conserv 68:63–68
Hoogesteijn R (2005) Manual on the problem of depredation caused Palmeira FBL, Barrella W (2007) Conflitos causados pela predação
by jaguars and pumas on cattle ranches. Wildlife Conservation de rebanhos domésticos por grandes felinos em comunidades
Society, New York, p 35 quilombolas na Mata Atlântica. Biota Neotrop 7:21–30
Inskip C, Zimmermann A (2009) Human–felid conflict: a review of Palmeira FBL, Crawshaw PG Jr, Haddad CM, Ferraz KMPMB,
patterns and priorities worldwide. Oryx 43:18–34 Verdade LM (2008) Cattle depredation by puma (Puma
Jackson P, Nowell K (1996) Problems and possible solutions in concolor) and jaguar (Panthera onca) in central-western Brazil.
management of felid predators. J Wildl Res 1:304–314 Biol Conserv 141:118–125
Jorge MLS, Galetti M, Ribeiro MC, Ferraz KMPMB (2013) Mammal Patterson BD, Kasiki SM, Selempo E, Kays RW (2004) Livestock
defaunation as surrogate of trophic cascades in a biodiversity predation by lions (Panthera leo) and others carnivores on
hotspot. Biol Conserv 163:49–57 ranches neighboring Tsavo National Parks, Kenya. Biol Conserv
Kissling DW, Fernández N, Paruelo JM (2009) Spatial risk assess- 119:507–516
ment of livestock exposure to pumas in Patagonia, Argentina. Paviolo A, Di Blanco YE, De Angelo CD, Di Bitetti MS (2009)
Ecography 32:807–817 Protection affects the abundance and activity patterns of pumas
Kolowski JM, Holekamp KE (2006) Spatial, temporal, and physical in the Atlantic forest. J Mamm 90:926–934
characteristics of livestock depredations by large carnivores Peterson MN, Birckhead JL, Leong K, Peterson MJ, Peterson TR
along a Kenyan border. Biol Conserv 128:529–541 (2010) Rearticulating the myth of human–wildlife conflict.
Li X, Buzzard P, Chen Y, Jiang X (2013) Patterns of livestock Conserv Lett 3:74–82
predation by carnivores: human–wildlife conflict in Northwest Polisar J, Maxit I, Scognamillo D, Farrell L, Sunquist ME, Eisenberg
Yunnan, China. Environ Manag 52:1334–1340 JF (2003) Jaguars, pumas, their prey base, and cattle ranching:
Linnell JDC, Aanes R, Swenson JE, Odden J, Smith ME (1997) ecological interpretations of a management problem. Biol
Translocation of carnivores as a method for managing problem Conserv 109:297–310
animals: a review. Biodivers Conserv 6:1245–1257 R Core Team (2013) R: a language and environment for statistical
Linnell JDC, Odden J, Smith ME, Aanes R, Swenson JE (1999) Large computing. R Foundation for Statistical Computing, Vienna,
carnivores that kill livestock: do ‘‘problem individuals’’ really Austria. http://www.R-project.org/
exists? Wildl Soc Bull 27:698–705 Ribeiro MC, Metzger JP, Martensen AC, Ponzoni FJ, Hirota MM
Maclennan SD, Groom RJ, Macdonald DW, Frank LG (2009) (2009) The Brazilian Atlantic forest: how much is left, and how
Evaluation of a compensation scheme to bring about pastoralist is the remaining forest distributed? Implications for conserva-
tolerance of lions. Biol Conserv 142:2419–2427 tion. Biol Conserv 142:1141–1153
Magioli M, Moreira MZ, Ferraz KMPMB, Miotto RA, Camargo PB, Rosas-Rosas OC, Bender LC, Valdez R (2008) Jaguar and puma
Rodrigues MG, Canhoto MCS, Setz EF (2014) Stable isotope predation on cattle calves in Northeastern Sonora, Mexico.
evidence of Puma concolor (Felidae) feeding patterns in Rangel Ecol Manag 61:554–560
agricultural landscapes in southeastern Brazil. Biotropica 0:1–10 Rosas-Rosas OC, Bender LC, Valdez R (2010) Habitat correlates of
Manfredo MJ, Vaske JJ, Teel TL (2003) The potential for conflict jaguar kill-sites of cattle in northeastern Sonora, Mexico. Human
index: a graphic approach to practical significance of human Wildl Interact 4:103–111
dimensions research. Human Dimens Wildl 8:219–228 Seibert P, Negreiros OC, Bueno RA, Emmerich W, Moura BV Netto,
Marchini S, Macdonald DW (2012) Predicting ranchers’ intention to Marcondes MAP, César SF, Guillamoun JR, Montagna RG,
kill jaguars: case studies in Amazonia and Pantanal. Biol Barreto RAA, Nogueira JCB, Garrido MAO, Mello LE Filho,
Conserv 147:213–221 Emmerich M, Mattos JR, Oliveira MC, Godoi A (1975) Plano de
Mazerolle MJ (2013) AICcmodavg: model selection and multimodel Manejo do Parque Estadual de Campos do Jordão. Boletim
inference based on (Q)AIC(c). R package version 1.33. http:// Técnico 19. Secretaria de Estado dos Negócios da Agricultura,
CRAN.R-project.org/package=AICcmodavg Coordenadoria da Pesquisa de Recursos Naturais, Instituto
Mazzolli M (1993) Ocorrência de Puma concolor (Linnaeus) Florestal, São Paulo
(Felidae, Carnivora) em áreas de vegetação remanescente de Shaw HG, Beier P, Culver M, Grigione M (2007) Puma field guide.
Santa Catarina, Brasil. Revta bras Zool 10:581–587 The Cougar Network. http://www.cougarnet.org
Mazzolli M (2010) Mosaics of exotic forest plantations and native Soto-Shoender JR, Giuliano WM (2011) Predation on livestock by
forests as habitat of pumas. Environ Manag 46:237–253 large carnivores in the tropical lowlands of Guatemala. Oryx
Mazzolli M, Graipel ME, Dunstone N (2002) Mountain lion 45:561–568
depredation in southern Brazil. Biol Conserv 105:43–51 Soto-Shoender JR, Main MB (2013) Differences in stakeholder
Miotto RA, Cervini M, Figueiredo MG, Begotti RA, Galetti PM Jr perceptions of the jaguar Panthera onca and puma Puma
(2011) Genetic diversity and population structure of pumas concolor in the tropical lowlands of Guatemala. Oryx
(Puma concolor) in southeastern Brazil: implications for 47:109–112

123
Environmental Management

Teichman KJ, Cristescu B, Nielsen SE (2013) Does sex matter? Wilson SM, Madel MJ, Mattson DJ, Graham JM, Merrill T (2006)
temporal and spatial patterns of cougar–human conflict in British Landscape conditions predisposing grizzly bears to conflicts on
Columbia. PLoS One 8:e74663 private agricultural lands in the western USA. Biol Conserv
Treves A, Wallace RB, Naughton-Treves L, Morales A (2006) Co- 130:47–59
managing human–wildlife conflicts: a review. Human Dimens Zar JH (2010) Biostatistical analysis, 5th edn. Pearson Prentice Hall,
Wildl 11:383–396 New Jersey
Vaske JJ, Needham MD, Newman P, Manfredo MJ, Petchenik J Zarco-González MM, Monroy-Vilchis O, Rodrı́guez-Soto C, Urios V
(2006) Potential for conflict index: hunters’ responses to chronic (2012) Spatial factors and management associated with livestock
wasting disease. Wildl Soc Bull 34:44–50 predations by Puma concolor in Central Mexico. Human Ecol.
Vaske JJ, Beaman J, Barreto H, Shelby LB (2010) An extension and doi:10.1007/s10745-012-9505-4
further validation of the potential for conflict index. Leis Sci Zimmermann A, Walpole MJ, Leader-Williams N (2005) Cattle
32:240–254 ranchers’ attitudes to conflicts with jaguars in the Pantanal of
Verdade LM, Campos CB (2004) How much is a puma worth? Brazil. Oryx 39:406–412
Economic compensation as an alternative for the conflict
between wildlife conservation and livestock production in
Brazil. Biota Neotrop 4:1–4

123

You might also like